Winning a new customer costs many times more than keeping an existing one — everybody knows that. Yet most small-shop points programmes keep nobody, because they stop at recording and never reach acting.
Step 0: Identify customers by something they already carry
Plastic cards get left at home. Nobody installs an app for a single café. What a customer always has and always remembers is a phone number. Start there, and only consider an app once the scale justifies it.
If identifying the customer slows the till, staff will skip it exactly when it is busy — that is, skip precisely the customers most worth recording. Entering the phone number must be on the payment screen itself, not behind another menu.
Choosing a mechanism: three kinds and where each fits
| Mechanism | Fits | Weakness |
|---|---|---|
| Points on spend, redeemed for rewards | Shops with varied products and order values | Points accrue slowly and customers lose patience |
| Visit count (buy 9, get 1 free) | Cafés, bubble tea, fast food | Easy to game without safeguards |
| Membership tiers with their own benefits | Spas, salons, fashion, higher-value services | Needs enough customers for tiers to mean anything |
The common mistake is running all three at once for completeness. Customers do not understand it, staff explain it wrongly, and in the end nobody uses it. Pick one, make it clear, revisit in six months.
What to measure
Cards issued is not a metric. These four say whether the programme is working:
- Share of orders with an identified customer — below 40% and everything analysed afterwards is unrepresentative.
- Return rate at 30/60/90 days — grouped by the month of first purchase so cohorts are comparable.
- Average order value of members versus walk-ins — if they are equal, the programme is just discounting people who were buying anyway.
- Share of points redeemed — points accrued but never redeemed mean the reward is unattractive or unreachable.
The most valuable part: customers about to leave
This is where the data creates real value, and where most shops do nothing. With identified sales data you can compute each customer’s normal purchase interval. A regular who buys coffee three times a week and has not appeared for two weeks is a signal — and it appears before the customer is lost.
The weekly list of "regulars unusually absent" is typically a few dozen people. One well-timed message with a small, specific offer wins far more of them back than a campaign blasted to the whole list.
Sending an offer to everyone is a discount. Sending it to the person about to leave is retention.
Three common mistakes
- 1Rules too complex. "Earn 10 points per 100,000đ, 500 points redeems a 50,000đ voucher valid on orders over 200,000đ" — nobody can compute that at the counter. Make it simple enough that staff can state it in one sentence.
- 2No expiry. Points that never expire sound generous but create an accumulating liability and give nobody a reason to come back soon.
- 3Not spendable at every branch. Points earned at one shop and unusable at another is the fastest way to destroy goodwill — see the shared-data section in the multi-branch article.
Want to talk specifics?
SealCore surveys at your premises and sends a fixed quote after the first session — including when the conclusion is that you do not need custom software.
